Partial Least Squares Regression Models for the Analysis of Kinase Signaling

Abstract

Partial least squares regression (PLSR) is a data-driven modeling approach that can be used to analyze multivariate relationships between kinase networks and cellular decisions or patient outcomes. In PLSR, a linear model relating an X matrix of dependent variables and a Y matrix of independent variables is generated by extracting the factors with the strongest covariation. While the identified relationship is correlative, PLSR models can be used to generate quantitative predictions for new conditions or perturbations to the network, allowing for mechanisms to be identified. This chapter will provide a brief explanation of PLSR and provide an instructive example to demonstrate the use of PLSR to analyze kinase signaling.

The design of high affinity human PD-1 mutants by using molecular dynamics simulations (MD)

“…Up to date, five antibody drugs targeting the PD-1/PD-L1 axis were approved by FDA. Many peptides and even small molecule modulators of the target have been under development [ 30 , 31 ]. Although the PD-1/PD-L1 related drugs have been successfully applied in clinic and several modulators showed bioactivities, the structural properties of hPD-1/PD-L1 and its binding mechanism in molecular level still needs to be studied.…”

The 7-year follow-up of the Hungarian BICAMS validation cohort implies that cognitive performance may improve in multiple sclerosis patients

“…PLS is able to analyze data with predictive variables that are numerous in number relative to the low number of observations and also highly correlatedwhich is the case with our clinical and cognitive data [24]. PLS creates orthogonal principal components predicting the Y variable through the linear combination of X variables [25]. The dependent variables in our evaluation were the total SDMT, BVMT-R, and CVLT-II scores at 5 years and 7 years, while the predictive factors were the clinical and sociodemographic data and FIS scores at baseline and the presence of DMT escalation during the observational period.…”

Análisis proteómico de la respuesta protectora mediada por el esteroide sintético Tibolona frente a un insulto lipotóxico en astrocitos humanos

“…Las técnicas más utilizadas incluyen el análisis de componentes principales y t-SNE (Giuliani, 2017;Oliveira et al, 2018). Otros enfoques de correlación, como el agrupamiento jerárquico y la regresión de mínimos cuadrados parciales, pueden evaluar las dependencias entre características o muestras de grupo y características en función de su similitud (Bourgeois and Kreeger, 2017;McLachlan et al, 2017;Si et al, 2014).…”